That's only partly true - it's ready to go when there's no showstopper bugs.
The issue here, Luca, is that you use ntop on one specific way on a small
set of platforms, and what works for you may (or may not) work for our
users.

As I read the lists, the showstoppers are

   8ZCKFWB --- Can Only Display One Local Net At A Time; Only Partial Stats
Collected For 2nd Net

   TPG2XT6 --- gzflush() failure


I've never been able to reproduce 8ZCKFWB, but it's persistent in his
environment.


And while the basic work-around for TPG2XT6 (5VHRELB et al) - disable
compressed html pages - is simple, it's a symptom of a serious problem of
memory corruption, and it's happening in Linux AND FreeBSD.  The fork()ed
child free fix has helped, but the problem is still there.  It's MUCH, MUCH
worse with glibc 2.3 (i.e. RH8+) so it will start to bite us as more distros
move to gcc 3.2/glibc 2.3.

The only way I can think to track down the other free() in child cases are
to turn on the diagnostic code in leaks.c and sift through the resulting
flood of problem reports.


As to the clueless, do what I do - send them a generic reply - it's precoded
into my mail client as a 'signature', with the pointers to the mailing
lists, search instructions and how to report problems.


As to who I support, well, fixing the things you break in the configure
files is a full time job (grin, duck & run).

Seriously, Luca, the configure files worked AND WERE TESTED in every
platform we support (seven distributions of Linux, FreeBSD 4.8 and 5.1 and
NetBSD 1.6.1) except the Mac (which I don't have access to) before your
latest changes.  Now they don't work anywhere EXCEPT the Mac.   If the Mac
platform is so different, how would you feel about splitting the configure
and makefiles into two a Mac version and a non-Mac version, like we used to
have for MinGW.


-----Burton



-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Luca
Deri
Sent: Thursday, August 07, 2003 3:00 AM
To: Ntop; Ntop-Dev
Subject: [Ntop] Re: Known problems - count down list for 2.3 - 1st cut


Burton,
when I worked for IBM my boss told me that a software is ready to be
released not when it's bug-free but when the open problems are under a
certain % and the advantage you get from the new release are greater
than the pain you have keeping the old release. On the other hand I
think it's time to put 2.2 into the trashbin because I don't have
enough time for telling people who contact me directly how to compile
rrd, why gdchart/png etc. have incompatibilities with versions mix &
match etc. So I'll see what I can do with the open issues then cut 2.3
and leave open problems to future releases.

Cheers, Luca

PS: If I fix all bugs who are you going to support then? :-)


_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to